Final Research Report WisDOT Project ID 0092-08-13

Report No. WHRP 10-01

FRICTION COEFFICIENTS FOR

STAINLESS STEEL/PTFE (TEFLON) BEARINGS

by

Josef C. Taylor and John F. Stanton

Department of Civil Engineering

University of Washington

Prepared for:

Wisconsin Department of Transportation.

Administrative Contact: Andrew Hanz.

Technical Contact: Travis MacDaniel.

January 2010

ii

Disclaimer

This research was funded through the Wisconsin Highway Research Program

by the Wisconsin Department of Transportation and the Federal Highway

Administration under Project 0092-08-13. The contents of this report reflect the views

of the authors who are responsible for the facts and accuracy of the data presented

herein. The contents do not necessarily reflect the official views of the Wisconsin

Department of Transportation or the Federal Highway Administration at the time of

publication.

This document is disseminated under the sponsorship of the Department of

Transportation in the interest of information exchange. The United States Government

assumes no liability for its contents or use thereof. This report does not constitute a

standard, specification or regulation.

The United States Government does not endorse products or manufacturers.

Trade and manufacturers’ names appear in this report only because they are considered

essential to the object of the document.

iii

iv

Technical Report Documentation Page 1. Report No. WHRP 10-01

2. Government Accession No

3. Recipient’s Catalog No

4. Title and Subtitle Friction Coefficients for Stainless Steel (PTFE) Teflon Bearings

5. Report Date: January 2010 6. Performing Organization Code Wisconsin Highway Research Program

7. Authors Josef C. Taylor, John F. Stanton

8. Performing Organization Report No.

9. Performing Organization Name and Address Department of Civil Engineering, Box 3527000, University of Washington, Seattle WA 98195

10. Work Unit No. (TRAIS) 11. Contract or Grant No. WisDOT SPR# 0092-08-13

12. Sponsoring Agency Name and Address Wisconsin Department of Transportation Division of Business Services Research Coordination Section 4802 Sheboygan Ave. Rm 104 Madison, WI 53707

13. Type of Report and Period Covered

Final Report, 2008-2010 14. Sponsoring Agency Code

15. Supplementary Notes 16. Abstract The research objectives are to determine the coefficients of friction of stainless steel-PTFE surfaces with two different surface finishes: #7 high-luster and #8 mirror, to provide an initial evaluation of the effects of higher friction on the forces in the structure, and to report the results to WisDOT in a way that can be easily and clearly understood, and implemented in bridge design practice through the Bureau of Structures.

17. Key Words Friction, PTFE, Teflon, Stainless steel, Sliding,

Bearing.

18. Distribution Statement No restriction. This document is available to the public through the National Technical Information Service 5285 Port Royal Road Springfield VA 22161

19. Security Classif.(of this report) Unclassified

19. Security Classif. (of this page): Unclassified

20. No. of Pages

21. Price

Form DOT F 1700.7 (8-72) Reproduction of completed page authorized

v

Executive Summary.

This report describes a study on sliding bridge bearing made from PTFE and stainless steel. Such bearings are commonly made from sheet PTFE and stainless steel polished to a #8 mirror finish. That surface finish is the only one for which the AASHTO LRFD Design Specifications provide design values of the friction coefficient. However, it can be expensive and difficult to obtain.

The objective of the work was to determine the suitability of stainless steel with a 2B surface finish in sliding bearings. 2B stainless steel is produced by cold rolling and is not polished. It is thus more readily available and less expensive, but it has a rougher finish.

A program of tests was undertaken to investigate the coefficient of friction and the wear characteristics of sliding bearings. Three stainless steel surface finishes were used: #8 mirror (as a reference), 2B, and a rough hot-rolled finish that was initially supplied by WisDOT. The results of the tests were analyzed and recommendations were prepared.

Friction between PTFE and a hard material such as stainless steel varies with many parameters, the most important of which are: surface finish, contact pressure, sliding speed, slide path and temperature. The first four of these were addressed in the tests; low temperature testing requires special equipment that lay outside the scope of the project.

The test results shared many characteristics with those found in previous studies. Static, or breakaway, friction is higher than sliding friction. The coefficient of friction is sensitive to contact pressure (unlike, for example, steel on steel, for which it is essentially independent of contact pressure), and increases at low pressure. It increases as sliding speed increases, although, within the range of sliding speeds expected in a non-seismic application, the sensitivity is not great.

By contrast, the effects of slide path were unexpected. For the mirror finish material, the coefficient of friction rose with increasing slide path, for the rough hot-rolled material, it fell, and for the 2B material it remained almost constant, even over one very long slide path test of three-quarters of a mile. At the end of each long slide-path test, the mirror finish material almost always displayed the highest friction coefficient. This result was counter-intuitive, but was consistent across essentially all tests with the three materials.

Wear of the PTFE was also measured, and was found to be very low for the 2B finish.

A general equation was developed from the test data with which to predict the friction coefficient as a function of surface finish, contact pressure, sliding speed and slide path.

Computations were done to estimate the slide path demand in a real bridge. It was found to vary greatly among bridges, and to depend on column stiffness, span length, superstructure type and details, and temperature profile.

2B surface finish stainless steel displayed stable and relatively low friction properties, based on which it was deemed to be a suitable alternative to #8 mirror finish, subject to the caveat that its performance characteristics at low temperature are unknown.

vi

Table of Contents ..................................................................................................................................................Page

List of Figures............................................................................................................................viii

List of Tables ...............................................................................................................................xi!

1 Introduction...............................................................................................................................1!

2 Previous Work...........................................................................................................................3!

2.1 Parameters that affect friction in PTFE-SS interfaces.................................................3!2.1.1 Stainless steel finish:..................................................................................................3!2.1.2 Contact Pressure:.......................................................................................................4!2.1.3 Eccentric loading.......................................................................................................4!2.1.4 Sliding speed:..............................................................................................................4!2.1.5 Grit, water, or other contaminants .........................................................................5!2.1.6 Slide-path: ...................................................................................................................5!2.1.7 Breakaway friction.....................................................................................................5!2.1.8 Creep of the PTFE....................................................................................................5!2.1.9 Temperature: ..............................................................................................................5!

3 Experimental Methods.............................................................................................................7!

3.1 Tests Performed................................................................................................................7!3.2 Test Specimens..................................................................................................................8!

3.2.1 PTFE and carrier.......................................................................................................8!3.2.2 Stainless Steel .............................................................................................................9!

3.3 Naming Convention and Test Matrix. .........................................................................11!3.4 Test Apparatus ................................................................................................................12!

3.4.1 Loading Apparatus. .................................................................................................12!3.4.1 Instrumentation .......................................................................................................15!

3.5 Test Procedure.................................................................................................................17!3.5.1 Loading History.......................................................................................................18!

4 Results ......................................................................................................................................21!

4.1 Observed Results ............................................................................................................21!4.1.1 Temperature .............................................................................................................21!4.1.2 Wear of PTFE..........................................................................................................21!4.1.3 PTFE Loss ...............................................................................................................24!

4.2 Recorded Results.............................................................................................................25!4.2.1 Load-displacement Curves.....................................................................................26!

4.3 Data Processing...............................................................................................................30!5 Analysis & Discussion of Results .........................................................................................35!

5.1 Introduction.....................................................................................................................35!5.2 Short Slide-Path, Multi-Speed, Tests............................................................................37!

vii

5.2.1 Effect of Sliding Speed........................................................................................... 37!5.2.2 Effect of Contact Pressure .................................................................................... 44!

5.3 Long Slide-path Tests..................................................................................................... 48!5.3.1 Problems with Data and Correction Methods.................................................... 48!5.3.2 Effect of Speed........................................................................................................ 53!5.3.3 Effect of Contact Pressure .................................................................................... 55!5.3.4 Effect of Slide-path................................................................................................. 59!5.3.5 Effect of Cyclic Displacement Amplitude........................................................... 68!

5.4 Warm-up Segment and Breakaway Friction ............................................................... 71!5.5 Extreme Cyclic Displacement Amplitude Test .......................................................... 74!5.6 Discussion of Results ..................................................................................................... 76!

6 Bearing Life .............................................................................................................................83!

6.1 Slide-path Expected in the Field................................................................................... 83!6.1.1 Thermal Movements............................................................................................... 83!6.1.2 Traffic Movements.................................................................................................. 86!

6.2 Bearing Slide-Path Capacity .......................................................................................... 91!7 Summary and Conclusions....................................................................................................94!

7.1 Summary .......................................................................................................................... 94!7.2 Conclusions ..................................................................................................................... 95!7.3 Recommendations. ......................................................................................................... 96!

7.3.1 Recommendations for Implementation............................................................... 96!7.3.2 Recommendations for Further Research............................................................. 97

Bibliography................................................................................................................................98

Appendix A, Problems During Testing..................................................................................99

Appendix B, Data Plots ..........................................................................................................103

viii

List of Figures Figure ................................................................................................................................... Page

Figure 3-1. Specimen set: PTFE carrier, PTFE disc, and stainless steel plate. ...................9!

Figure 3-2. Mirror specimen before test (M1.45.250.100) .....................................................9!

Figure 3-3. 2B stainless steel plate before testing. .................................................................10!

Figure 3-4. Rough specimen before testing (R1.30.250.100)...............................................10!

Figure 3-5. Schematic drawing of the testing rig. ..................................................................13!

Figure 3-6. Test rig; four vertical rams (center), and horizontal actuator (left). ...............14!

Figure 3-7. Showing the specimen carriers and pots 4 and 5 ..............................................17!

Figure 3-8. Schematic of testing sequence .............................................................................19!

Figure 4-1. mirror polished specimen after test M1.45.250.100 .........................................22!

Figure 4-2. 2B rolled specimen after B2.45.250.100 test......................................................22!

Figure 4-3. Rough rolled specimen after R1.45.250.100 test...............................................23!

Figure 4-4. Idealized force displacement curve. ....................................................................26!

Figure 4-5. Typical load-displacement curve for #8 mirror polish test .............................27!

Figure 4-6. Typical load-displacement curve for 2B rolled plate test.................................28!

Figure 4-7. Typical load-displacement curve for rough rolled plate test ...........................29!

Figure 4-8. Load displacement curve for B2.15.999.100 test showing individual datapoints..........................................................................................................................................30!

Figure 4-9. Sparse data during the B2.15.384.005 does not adequately represent the cycle..........................................................................................................................................31!

Figure 4-10. Several deficient cycles superimposed reveal a better picture. ......................32!

Figure 5-1. Example data showing data increasing or decreasing with slidepath.............37!

Figure 5-2. Multi-speed test, 1500 psi contact pressure, Long slide-path test at 2.5 in/min, ±1.0” stroke ..................................................................................................................38!

Figure 5-3. Multi-speed test, 3000 psi contact pressure, Long slide-path test at 2.5 in/min, ±1.0” stroke ..................................................................................................................39!

Figure 5-4. Multi-speed test, 4500 psi contact pressure, Long slide-path test at 2.5 in/min, ±1.0” stroke ..................................................................................................................39!

Figure 5-5. Multi-speed test at 1500 psi. Ratio of B2 friction to M1 friction vs. speed before and after the long slide-path test at 2.5in/min and ±1.0” stroke .............41!

Figure 5-6. Multi-speed test at 3000 psi. Ratio of B2 friction to M1 friction vs. speed before and after the long slide-path test at 2.5in/min and ±1.0” stroke .............41!

ix

Figure 5-7. Multi-speed test at 4500 psi. Ratio of B2 friction to M1 friction vs. speed before and after the long slide-path test at 2.5in/min and ±1.0” stroke .............42!

Figure 5-8. All multi-speed tests, before and after a long slide-path test, friction amplification factor normalized to 1 at 2.5in/min for each test. ..........................43!

Figure 5-9. CoF vs. contact pressure for 0.08 in/min, before and after long slide-path test at 2.5 in/min and ±1.0” stroke ..................................................................................44!

Figure 5-10. CoF vs. contact pressure for 0.64 in/min, before and after long slide-path test at 2.5 in/min and ±1.0” stroke ..................................................................................44!

Figure 5-11. CoF vs. contact pressure for 2.5 in/min, before and after long slide-path test at 2.5 in/min and ±1.0” stroke ..................................................................................45!

Figure 5-12. Ratio of B2 friction to M1 friction vs. contact pressure before and after the long slide-path test for 0.08 in/min...........................................................................46!

Figure 5-13. Ratio of B2 friction to M1 friction vs. contact pressure before and after the long slide-path test for 0.64 in/min...........................................................................46!

Figure 5-14. Ratio of B2 friction to M1 friction vs. contact pressure before and after the long slide-path test for 2.5 in/min.............................................................................47!

Figure 5-15. Extreme long slide-path test, showing correlation of contact pressure and temperature to friction. Sliding speed was 10.0 in/min, at 1500 psi nominal contact pressure..........................................................................................................................49!

Figure 5-16. CoF and linear function of contact pressure. Sliding speed was 10.0 in/min, at 1500 psi nominal contact pressure. .......................................................................50!

Figure 5-17. Extreme long slide-path test, showing good fit of pressure-predicted CoF, and corrected friction. Sliding speed was 10.0 in/min, at 1500 psi nominal contact pressure..........................................................................................................................51!

Figure 5-18. AASHTO design CoFs for #8 mirror polish..................................................52!

Figure 5-19. CoF for B2 long slide-path tests at ±1.0” and 1500 psi, all speeds, and multi-speed test for B2.15.250.100.......................................................................................53!

Figure 5-20. CoF for B2 long slide-path tests at ±0.05” and 1500 psi, all speeds. ..........54!

Figure 5-21. Ratio of muB2 to muM1 during long slide-path tests at 2.5in/min, ±1" stroke and 3.84in/min, ±.05" stroke.....................................................................................55!

Figure 5-22. Concatenation of CoF vs. slide-path curves. Rough plate, 2.5 in/min. ......57!

Figure 5-23. CoF vs. Pressure for the three 2.5 in/min tests done on each of the finishes of SS, showing trend-lines and 95% Confidence Interval. (CIs are the higher lines)........................................................................................................................................58!

Figure 5-24. Coefficient of friction vs. path for different contact pressures on #8 mirror polish specimens at 2.5 in/min ..................................................................................60!

x

Figure 5-25. Coefficient of friction vs. path for different contact pressures on 2B Rolled plates at 2.5 in/min ......................................................................................................61!

Figure 5-26. Coefficient of friction vs. path for different contact pressures on rough rolled specimens at 2.5 in/min ..............................................................................................61!

Figure 5-27. CoF v. Path for all finishes during long slide-path test (4500 psi, 2.5 in/min, ±1" stroke) ....................................................................................................................63!

Figure 5-28. CoF v. Path for all specimens during long slide-path test (3000 psi, 2.5 in/min, ±1" stroke)......................................................................................................63!

Figure 5-29. CoF v. Path for all specimens during long slide-path test (1500 psi, 2.5 in/min, ±1" stroke)......................................................................................................64!

Figure 5-30. CoF v. Path for all specimens during high speed, low displacement test (1500 psi, 3.84 in/min, ±.05" stroke) ...................................................................................65!

Figure 5-31. Normalized CoF vs. Path for tests done at 2.5 in/min and ±1.0” cyclic displacement..................................................................................................................67!

Figure 5-32. Normalized CoF for extreme long path test at 1500 psi, 10 in/min and ±1.0" cyclic displacement.......................................................................................................67!

Figure 5-33. Long slide-path test on 2B Rolled Plates at 1500 psi and 2.5in/min, different displacement amplitudes. ............................................................................................69!

Figure 5-34. Long slide-path test on 2B Rolled Plates at 1500 psi and 3.84in/min, different displacement amplitudes. ............................................................................................69!

Figure 5-35. Long slide-path test on 2B Rolled Plates at 1500 psi and 10in/min, different displacement amplitudes. ............................................................................................70!

Figure 5-36. CoF vs. path during warmup cycles for #8 mirror polish specimens. Sliding speed is 2.5 in/min and stroke is ±1.0" except for M1.15.384.005, where Sliding speed is 3.84 in/min and stroke is ±0.05"................................................................73!

Figure 5-37. CoF vs. path during warmup cycles for 2B rolled specimens. Sliding speed is 2.5 in/min and stroke is ±1.0" except for M1.15.384.005, where Sliding speed is 3.84 in/min and stroke is ±0.05”...............................................................................73!

Figure 5-38. CoF vs. path during warmup cycles for rough rolled specimens. Sliding speed is 2.5 in/min and stroke is ±1.0" except for M1.15.384.005, where Sliding speed is 3.84 in/min and stroke is ±0.05”...............................................................................74!

Figure 5-39. Extreme cyclic displacement amplitude test, showing force and contact pressure. .........................................................................................................................75!

Figure 6-1. Girder end rotation causing bearing slip ............................................................87!

xi

List of Tables Table ....................................................................................................................................Page

Table 3-1. Profilometer results for the three finishes of SS used. ......................................11!

Table 3-2. Test Matrix...............................................................................................................12!

Table 3-3. Sequence of sliding speed tests during multi-speed tests. .................................19!

Table 4-1. Polishing effect of PTFE Stainless Steel interfaces. ..........................................23!

Table 4-2. Measurement of PTFE loss during the tests. .....................................................25!

Table 5-1. Coefficients for the amplification function of sliding speed ............................43!

Table 5-2. Pressure correction coefficients for the three finishes of SS for 2.5 in/min sliding speed..................................................................................................................58!

Table 5-3. Coefficients for the amplification function of sliding speed and contact pressure........................................................................................................................................59!

Table 5-4. CoF at reference conditions (3000 psi contact pressure, 2.5 in/min, zero slide-path) ...............................................................................................................................66!

Table 5-5. Breakaway CoF from cycles 6 through 12 of the warm-up test, all at 2.5 in/min, ±1.0" displacement ......................................................................................................72!

Table 5-6. Comparison of CoF between AASHTO and this study ...................................77!

Table 5-7. CoF at reference conditions (3000 psi, 2.5 in/min, ±1.0") ..............................78!

Table 5-8. Coefficients for determining pressure..................................................................78!

Table 5-9. Suggested CoF at 2.5 in/min and worst-case slide-path...................................79!

xii

xiii

Acknowledgements The financial support of the Wisconsin Department of Transportation, and the technical assistance of Travis MacDaniel and Finn Hubbard in particular, is gratefully acknowledged. The laboratory work could not have been completed without the help of Vince Chaijaroen, the Structures Laboratory manager, whose assistance is much appreciated.

1

1 Introduction Bridges, especially those exposed to wide temperature variations or subjected to heavy

truck traffic, require bearings that can accommodate the movement of the bridge

superstructure while transmitting the least possible force to the substructure. Bearings

must perform their duties with minimal maintenance over as long a design life as possible,

due to the difficulty of replacing them.

Bearings accommodate movement by one of several different means. Elastomeric bearings

deform, steel rocker and roller bearings roll in rigid body rotation, and in sliding bearings

one material slides on another. This report concerns a particular type of sliding bearing,

made from PTFE (PolyTetraFluoroEthylene, known more commonly by the DuPont™

trade name Teflon™) and polished stainless steel. Teflon is chemically very inert, and was

originally developed as an electrical insulator. However its low friction characteristics and

its chemical inertness have made it popular in many commercial and residential

applications, such as non-stick cookware, in addition to its use in bridge bearings.

Several types of PTFE are used in bridge bearings. In the US, flat sliding bearings are

usually made with solid sheet PTFE. That material is most commonly pure PTFE, but in

the past it has been reinforced with glass or other fibers to inhibit creep. That practice has

largely been discontinued, because the fibers scratch the stainless steel and increase the

friction. In Europe and Canada, the sheet PTFE is frequently dimpled and lubricated with

grease. Lubrication decreases the friction to very low levels (to less than 0.3% in some

cases) and the dimples act as reservoirs, without which the grease would soon be squeezed

out from between the PTFE and the stainless steel. In spherical sliding bearings, the

PTFE needs to be formed into a spherical shape, and doing so is difficult. Therefore a

woven material is often used that is made from PTFE thread backed by reinforcing

threads made from other more robust materials, such as glass fibers.

Bearing design in the USA is usually controlled by the requirements of the AASHTO

(American Association of State Highway and Transportation Officials) LRFD Bridge

Design Specifications (Section 14.7.2). For sliding bearings, they give values of the friction

2

coefficient to be used for design. Separate values are given for different types of PTFE,

but for the stainless steel only one surface finish, namely a #8 mirror polish, is specified.

(That finish is highly polished and was originally developed to create mirrors for

circumstances in which glass might be subject to breakage. Hence its name). It has given

good service and leads to a low coefficient of friction, but is sensitive to scratches and

other surface imperfections, which can be cause for rejection. The polishing process is

also time consuming and may lead to difficulties in obtaining the material within

reasonable lead times, especially after a last minute rejection.

Interest therefore exists in determining whether other, less highly polished but more

readily available, stainless steel finishes might provide adequate performance. A finish

known as 2B is widely available, and is a potential alternative. It is produced by cold

rolling, using highly polished rollers. No further processing is used, and so the material is

relatively economical and readily available. It is widely used in the food-processing

industry

The report documents tests performed to determine the friction characteristics of sheet

PTFE sliding against three different stainless steel finishes. The three finishes were: #8

mirror polished, 2B rolled, and a rolled stainless steel with a relatively rough surface. The

tests were performed in the Structural Engineering Laboratory of the Civil Engineering

Department of the University of Washington.

3

2 Previous Work Significant research has been done on PTFE-Stainless steel friction interfaces. Papers have

discussed many different factors that determine the friction characteristics of these

interfaces, including the PTFE itself. In addition to plain (unfilled) PTFE, several other

materials which use PTFE have been studied. Jordain T. Rheault, (1992) focused on the

behavior of different types of PTFE (dimpled-lubricated, woven, unfilled, 15% and 25%

glass filled). Lubricated PTFE has much lower friction, but is vulnerable to contamination

(Campbell and Manning 1993). Woven PTFE can be draped over compound curves, such

as spherical bearings. For this material, static friction is closer to sliding friction. Fiber

glass impregnated PTFE is used when wear and creep are a concern, or to allow higher

contact pressure, but also has higher friction (Rheault, 1992).

2.1 Parameters that affect friction in PTFE-SS interfaces

Campbell and Kong (1989) identified the following 14 parameters as influencing the

friction coefficient:

• Type of PTFE • Dimpling and lubrication • Specimen size • Attachment of the PTFE to the backing plate • Roughness of the mating surface • Contact pressure • Eccentric loading • Speed of travel • Length of the travel path • Load and travel history • Surface contamination • Wear • Creep • Temperature

2.1.1 Stainless steel finish: The coefficient of friction is generally believed to increase with increasing surface

roughness. Surface rougness is measured with a profilometer. Several scales are used. The

4

most common is Ra, the arithmetic average of absolute distance from the mean. Campbell

and Kong tested PTFE-SS interfaces with SS surface roughness of 1.18 and 13.4 !in Ra.

These finishes are slightly smoother than #8 mirror polish, and slightly rougher than 2B

rolled finish, respectively. These tests were conducted using lubricated PTFE

2.1.2 Contact Pressure: The coefficient of friction decreases with higher contact pressure. That is, doubling the load

on an interface will yield a less than double increase in friction force. This is why drag

racers, whose success depends on getting as much friction force out of their tires as

possible, use very large, slick tires, and inflate them only enough to hold up the car.

NCHRP Report 432, High-Load Multi-Rotational Bridge Bearings (Stanton et al. 1999)

describes the results from extensive testing of #8 mirror polished SS-PTFE interfaces at

contact pressures from 500 to 6000 psi. The report proposed that CoF is approximated as

decreasing linearly with increasing contact pressure below 3000 psi, and being constant at

higher pressure The tabulated friction values published in AASHTO Bridge Design

Specification, (18.1.5.2.6 2009) come from this report.

2.1.3 Eccentric loading An eccentric load causes variations in contact pressure over the surface, which can cause

changes in friction coefficient. Since higher pressures produce less change in CoF than

lower pressures, eccentricities tend to mean that friction is increased. Local high pressure

also causes increased wear. Dimples in PTFE may be damaged by increased pressure, or

the PTFE may be squeezed out of position.

2.1.4 Sliding speed: Coefficient of friction increases with sliding speed, but approaches a constant value

(Campbell and Kong 1989), (Stanton et al. 1999), (Constantinou 1999). The speed at

which it reaches a constant value differs for different materials and surface finishes. For

conventional unfilled PTFE, and #8 mirror polish SS, that critical speed is much higher

than any speed likely to occur in practice. High speed sliding is of interest primarily in

seismic isolation systems.

5

2.1.5 Grit, water, or other contaminants Contamination of the sliding surface increases the CoF. This is especially true for

lubricated PTFE, as the lubrication stops working when contaminated. Campbell and

Manning (1993) introduced contamination to dry and lubricated unfilled PTFE in the

form of portland cement powder. They found that, though contamination can be

introduced before the bearing is put together, it is difficult to introduce contamination

once the bearing is in service.

2.1.6 Slide-path: CoF changes with travel path length. The first few cycles see dramatically decreasing CoF,

especially static friction during reversals. Subsequently, CoF may increase slowly with

greater travel path, as is the case for #8 mirror polish SS mating surfaces. This is believed

to be caused by transfer of PTFE to the mating surface, which causes the interface to

mimic a PTFE-PTFE interface, which has a higher CoF.

2.1.7 Breakaway friction CoF is significantly higher under static conditions than dynamic, such that reversals of

motion during cyclic action cause peaks in friction force.

2.1.8 Creep of the PTFE PTFE exhibits creep behavior under large sustained loads, but normal conditions in bridge

bearings will not cause sufficient creep to affect performance. The most common

manifestation of creep occurs when high local pressure causes the PTFE to flow out of

the recess that is meant to contain it.

2.1.9 Temperature: The coefficient of friction is not affected at moderate temperatures, but increases with

decreasing temperature below 32ºF. (Campbell and Kong 1989), Also, low temperatures

cause increased wear (Stanton et al. 1999). Constantinou, et al. (1999) introduced property

modification factors to allow for the many ways in which PTFE-SS interfaces may

perform worse than is documented, including contamination of the interface, long travel

paths, and low temperatures. The factor for low temperatures ranges from 1.0 at 20°C to

2.0 at -50°C

6

7

3 Experimental Methods A series of experiments was performed to determine the effects of stainless steel surface

finish on friction coefficient. Friction associated with a 2B rolled finish was of greatest

interest, so the experiments focused on comparing the behavior of a 2B rolled finish with

that of a conventional #8 mirror polished stainless steel. A third material ( a relatively

rough, hot rolled plate) was originally provided. It was tested as well to provide a broad

range of data, even though it is unlikely to be chosen in practice.

3.1 Tests Performed

Eighteen specimen pairs were tested in all. Each specimen was subjected to a series of

tests, at different sliding speeds. The same test program was applied to all specimens1. The

tests were distinguished by the following features:

• Stainless steel surface finish (#8 mirror polished, 2B rolled, and rough rolled).

• Contact pressure (1500, 3000, 4500 psi).

• Sliding speed for long slide-path test (2.5, 3.84, or 10 in/min)

• Cyclic displacement amplitude during long slide path test (±0.05”, ±0.25”, or ±1”)

The stainless steel surface finishes were chosen because:

• #8 mirror polish is the conventionally used surface finish (AASHTO 2009,

Campbell and Kong 1989, Constantinou 1999, Stanton et al 1999) so was needed

to correlate the data from this study with the findings of other research.

• 2B rolled finish is a possible alternative to mirror that is inexpensive and readily

available.

• Rough rolled plates were sent by mistake, but were tested to investigate the effects

of a very rough surface finish.

1 Except R1.45, which was the first specimen tested. This is explained in 3.5.1

8

The contact pressures were chosen to reflect practice. WisDOT supplied several plans for

bridges which use PTFE-SS bearings (WisDOT structures B42-111, B20-168, B51-113).

The bearings are sized such that the contact pressure under service loading (dead and live)

would be about 1500 psi. Contact pressures up to 4500 psi are allowed by AASHTO.

Eccentricities and non-uniform girder loading can cause contact pressures larger than

planned.

NHCRP 432 (Stanton et al 1999) used a sliding speed of 2.5 in/min, and AASHTO

18.1.5.2.6 (LRFD Bridge Construction Specifications 2009) uses this speed in the criteria

for the testing of bearings, and so it was chosen as a starting point. Other sliding speeds

were chosen in the range of thermal movements and traffic vibrations. Thermal

movements occur at very slow speeds: a 100’ bridge (with a coefficient of thermal

expansion of ~5.5E-6 in/in/°F) warming by 100°F will change length by 0.66 inches over

the course of 12 hours, or about 0.001 in/min. Tests conducted at this speed would take

too long, so a minimum speed of 0.08 in/min was chosen for expedience. The highest

speed (at least in the standard tests) of 3.84 in/min was taken from an example bridge

calculation done to reflect traffic loading. Unfortunately, it was later found to be slower

than the true value. The 10 in/min speed, used in one special test at the end of the

program, was chosen so that a very large slide-path could be applied over the course of a

few days, as well as to better reflect motion due to traffic vibrations.

Displacements were chosen to relate to AASHTO 18.1.5.2.6 (LRFD Bridge Construction

Specifications 2009), which specifies a displacement of ±1.0” for testing. The amplitude

±0.05” was chosen to study the effects of very small, frequent displacements that might

occur because of traffic vibrations, and ±0.25” was chosen as an intermediate value.

3.2 Test Specimens

3.2.1 PTFE and carrier Each PTFE sample consisted of a 3” diameter, 1/8” thick circular piece of PTFE,

recessed into a 0.813”x5”x7” carbon steel plate. The plates were surface ground to

minimize any irregularities of thickness and to facilitate the measurement of PTFE loss.

The circular recess was machined to accept the PTFE discs, which were cut with an

9

abrasive water jet machine. The PTFE discs were then bonded into their recesses with

Caseway B-500 Etched Teflon Adhesive. The bottom surface of the PTFE had been

etched by the supplier. A schematic drawing of a typical specimen is shown in Figure 3-1.

Assembled specimens are shown in Figure 3-2 and Figure 3-4.

Figure 3-1. Specimen set: PTFE carrier, PTFE disc, and stainless steel plate.

3.2.2 Stainless Steel The stainless steel specimens were 5”x7”, and most were supplied as 0.5” thick plate.

Three finishes of stainless steel were used. #8 mirror polished finish is the currently

specified finish for PTFE-SS bearings, so was necessary for comparison and to validate

the results of this study. It is produced by surface grinding with progressively finer

grinding and buffing wheels. Figure 3-2 shows an example of the mirror polish plates and

their corresponding PTFE specimens.

Figure 3-2. Mirror specimen before test (M1.45.250.100)

2B rolled finish is an economical, readily available finish which does not require polishing.

It is produced by running the material through a set of highly polished rollers. The

samples provided were thin sheets (approximately 11 gage), and so were spot welded to

10

"” backing plates. These welds lie outside of the travel path, and the contact area was

protected during welding. Figure shows an example of a 2B rolled plate.

Figure 3-3. 2B stainless steel plate before testing.

Rough rolled finish is a common finish for SS plate, known as No. 1 finish, that was sent

by mistake, but provided interesting information about rougher finishes. It is produced by

hot rolling, followed by annealing, pickling, and passivation to remove mill-scale and

prevent corrosion. Figure 3-4 shows an example of the rough rolled plates and their

corresponding PTFE specimens.

Figure 3-4. Rough specimen before testing (R1.30.250.100)

The surface roughness of each plate was measured using a digital profilometer. The results

are shown in Table 3-1:

11

Table 3-1. Profilometer results for the three finishes of SS used.

Finish Average Surface Roughness (!in Ra)

#8 mirror polish 4 (stdev = 0.8)

2B rolled 6.5 (stdev = 1.2)

Rough rolled 190 (stdev = 20)

3.3 Naming Convention and Test Matrix.

The tests were named according to their surface finishes and the tests performed on them.

The names consist of the following terms, separated by periods:

• Finish & Batch – M for #8 mirror polish, R for rough or as-rolled, B for 2B

rolled; The first batch was the set of mirror and rough plates provided at the

beginning of testing. The second batch was all of the 2B rolled plates, purchased

from a Wisconsin materials supplier.

• Contact pressure – the bearing pressure on the PTFE interface in psi/100.

• Sliding speed – sliding speed during the long slide-path test in in/min x100.

• Displacement – one-way displacement during the long slide-path test in inches

x100

For example, the first specimen tested, a rough rolled interface from the Batch 1, which

was subjected to 4500 psi contact pressure and a long slide-path test at 2.5 inches per

minute and ±1.0” cyclic displacement amplitude is called R1.45.250.100. A matrix of the

tests and their names is shown in Table 3-2.

12

Table 3-2. Test Matrix Sliding Speed: 3.84 ipm 10 ipm

Pressure: 4500 psi 3000 psi 1500 psi 1500 psi 1500 psi

#8 Mirror Polished±1.0” M1.45.250.100 M1.30.250.100 M1.15.250.100 • •±0.05” • • • M1.15.384.005 •

2B Rolled ±1.0” B2.45.250.100 B2.30.250.100 B2.15.250.100 B2.15.384.100 B2.15.999.100±0.25” • • B2.15.250.025 B2.15.384.025 •±0.05” • • B2.15.250.005 B2.15.384.005 B2.15.999.005

Rough Rolled±1.0” R1.45.250.100 R1.30.250.100 R1.15.250.100 • •±0.05” • • • R1.15.384.005 •

2.5 ipm

3.4 Test Apparatus

3.4.1 Loading Apparatus. A test rig that had been built for a previous project was adapted for use with the PTFE-

stainless steel friction specimens. It is shown schematically in Figure 3-5. It consists of a

fixed reaction frame, a set of static rams to provide the normal force across the sliding

interface, a servo-controlled actuator that imposes the sliding displacements, and a

specimen carrier assembly. The actuator body is attached rigidly to the test frame, with no

allowance for pivot or twist. The moving end of the actuator is attached to the moving

specimen carrier, shown in detail in Figure 3-5. The PTFE specimens are held by the static

plates, which are bolted to the channels which form the uprights of the frame. Hydraulic

rams, whose rods pass through the static plates, but run by the moving plates, provide the

compressive force on the bearings. They are positioned so that the PTFE specimen is

located at the centroid of the compressive force.

.

13

Figure 3-5. Schematic drawing of the testing rig.

The specimens are held in place by #” keeper plates, shown in the detail of Figure 3-5.

These are bolted to tapped holes in the static plates, or through the moving plate.

Figure 3-6 shows the static plates that hold the PTFE specimens and the four rams that

provide the vertical force. The MTS 110k actuator that applied the shear force can be seen

on the left.

14

Figure 3-6. Test rig; four vertical rams (center), and horizontal actuator (left).

The four rams were connected via a series of T-connections and a check-valve to a

hydraulic pump. This arrangement provided a relatively constant bearing pressure at the

two sliding interfaces. Proper alignment and pressure were achieved by first setting the

hydraulic pressure in the vertical rams to the desired contact pressure, aligning and bolting

down the static plates, then making any final adjustment of the ram pressure.

If there is significant change in the oil pressure in the normal force rams, the static plates

would bend slightly, taking some of the force. This would mean that the contact pressure

would not be accurately predicted by the oil pressure. If the bolts slip at a small enough

force, the oil pressure would again reflect the contact pressure. A brief calculation was

performed to determine the change in contact pressure that would cause bolt slip. Based

on the torque on the bolts, and the eccentricity of the load, the bolts could withstand a

force of 7.8 kips, or a change in contact pressure of 1100 psi at the specimen. Since there

were no variations in contact pressure this great, the bolts are predicted not to have

slipped. On no occasion were they heard to slip. This means that any difference between

MTS Load Cell

Hydraulic Rams

15

the oil pressure and the contact pressure is taken up in bending of the static plates. An

approximate calculation was conducted to determine the relative stiffnesses of the plates

in bending (as cantilevers) and the PTFE in compression. The bolted connection was

assumed to provide the plates with full fixity against rotation, but the loadpath through the

PTFE was found to be 5 times as stiff as that through the plates, so 5/6 of any change in

ram force would be transmitted to the PTFE. It is likely that the bolted connection

permitted some local flexibility, in which case the proportion of any change in force

transmitted through the PTFE would be higher than 5/6.

Because the deformation in the static plates was elastic, the PTFE contact pressure would

remain a linear function of the oil pressure in the rams. This is important because later

analysis (the correction of CoF data based on variation of pressure, see Section 5.3.1)

depends on the oil pressure being related to the PTFE contact pressure by a linear

function.

3.4.1 Instrumentation The compressive and shear loads on the bearings were measured. The compressive load

was originally measured using one load cell on each of the four threaded rods (as shown in

Figure 3-5). However, the load cells had been designed and fabricated in-house for a

previous project, and were found in the first test (R1.45.250.100) to be unreliable. In later

tests, hydraulic pressure in the rams was measured using a pressure cell. This provided a

much more consistent reading. The load cells remained as a backup.

The shear load was measured using an MTS 110k load cell, model no 661.23E-01, located

between the MTS actuator and the specimen carrier assembly.

Displacements were measured using potentiometers. Figure 3-7 shows the locations of the

vertical potentiometers. These were intended to record any changes in thickness in the

bearings and any tilting due to the loading. Sliding displacements were measured by the

LVDT in the MTS actuator. During the first test, R1.45.250.100, in which loads were the

highest, the motions of the moving plates and the static plates were measured relative to a

common reference (the end channels of the frame to which the static plates are bolted).

This was intended to detect any elastic deformation of the rig. The motion of the moving

16

plate relative to the static plates matched the readings of the LVDT to an RMS error of

8.2E-3 inches during this test. As this was the test with the largest actuator load (high

contact pressure, rough rolled, low slide-path), this error was deemed acceptable. The

string pots were needed for another test, so were removed.

Later tests were carried out with a cyclic displacement amplitude of ±0.05”. For this

displacement, and RMS error of 0.0082” would represent 16% of the measured value.

Because the displacement amplitude was used in computing the friction coefficient (see

Section 4.3), errors in measuring the displacement are important. However, the error of

0.0082” was measured with the highest contact pressure (4500 psi), with the roughest

stainless steel, so the forces on the rig, which would cause rig deformation, were at their

largest. All of the ±0.05” tests were run at 1500 psi, thus, if friction were independent of

pressure, and the same for all surface finishes, the error might be expected to be 16% x

1500psi/4500 psi = 5%. Thus if a friction coefficient was measured to be 5.7%, its true

value could be between 0.95 and 1.05 times 5.7%. The error in friction coefficient caused

by the displacement measurement is thus likely to be smaller than the inherent scatter in

the data.

17

Figure 3-7. Showing the specimen carriers and pots 4 and 5

The static plates were bolted securely to the frame of the test rig and so were not expected

to move appreciably during testing. However, potentiometers were used to detect any

unintended movements; in the vertical direction the movements were measured relative to

each other, and in the longitudinal direction they were measured relative to the test frame.

The maximum measured change during the first test, R1.45.250.100, was 6.3x10-4 inches,

much smaller than even the smallest cyclic displacement amplitudes. Movement during

other tests was likely smaller, as this test saw the largest actuator forces of any test.

3.5 Test Procedure

The test specimens were cleaned, measured and photographed before being loaded into

the test rig. The hydraulic rams were then set to the desired contact pressure, compressing

the specimen. The thick static plates were aligned parallel to the moving plate and to each

other with dial gages. The static plates were then bolted to the reaction frame so that

horizontal movements of the bearings would not cause movement of the plates. The

Pot 4

Pot 5

Spec imens

18

instrumentation and recording was activated, and the servo-controlled horizontal actuator

was used to implement the test program. The test was performed over the course of the

next 24-72 hours, after which the specimens were removed from the rig, photographed,

and measured. Any loose material was brushed off the surface of the PTFE before its

thickness was measured

3.5.1 Loading History The horizontal load was applied under displacement control using a triangular ramp

function. This allowed a constant sliding speed throughout the cycle, and a sharp change

in direction at the reversal point.

The loading program finally adopted is illustrated in Figure 3-8. It consisted of:

• 20 cycles of “warm-up” movement at 2.5 in/min and ±1.0” displacement.

• A multi-speed test. This consisted of six sets of 3 cyclic displacements in which

each set occurred at a different sliding speed, and a displacement of ±1.0”. The

sliding speeds varied in a geometric progression from 0.08 to 2.5 in/min. The sets

were conducted in alternating order to enable the effects of sliding speed to be

distinguished from the effects of slide-path, since each test added to the total slide-

path experienced by that specimen. This sequence is shown in Table 3-3.

• A long slide-path wear test. 1600 inches of slide-path were applied at a constant

sliding speed and displacement amplitude to determine the way in which the

friction changed with wear, as characterized by the slide-path.

• A second multi-speed test, identical to the first, to determine the relationship

between friction coefficient and sliding speed after the wear test.

19

Table 3-3. Sequence of sliding speed tests during multi-speed tests.

Sliding Speed(in/min)

1 2.52 0.083 1.254 0.165 0.646 0.32

Test Sequence

Figure 3-8. Schematic of testing sequence

The warm-up cycles were selected because, especially with the rough stainless steel, the

friction changed significantly during the first few cycles. The friction in those first cycles

was considered to be less relevant than the subsequent values because:

• Bearings in a real bridge are likely to experience some movement during

construction, at which time the full vertical load is not present. During those

movements, some PTFE will be transferred to the stainless steel and the friction

will change. Even though the friction coefficient may be higher under those

circumstances, the friction force, which is the measure of potential damage to the

20

adjoining elements, is likely not to be critical because the normal force is smaller at

that time.

• To conduct a meaningful comparison between different sliding materials, stable

friction values are needed. These occur only after a few cycles of displacement

have been applied.

• Article 18.1.5.2.6 of the AASHTO Bridge Construction Specification (2009)

recommends that the breakaway friction from the sixth through twelfth cycles at

2.5 in/min, ±1.0” displacement should be used for design. The warm-up

procedure allows the measurement of these values.

This protocol was followed in nearly all of the tests, with the exception of the first

specimen pair, R1.45.250.100. In that test, the warm-up was omitted, and the first of the

multi-speed tests was done in descending order of sliding speed, and included an extra

speed, 0.04 in/min. The long slide-path test was performed as usual, and then a special

large displacement test was performed, to explore the engagement of new, clean stainless

steel. For further explanation of test R1.45.250.100, see Section 5.5. After this test, the

(rough) plates were flipped over and tested on their opposite sides. A shorter long slide-

path test was performed, followed by the speed test as performed in later tests.

In M1.45.250.100, large sideways displacements were detected which affected the friction.

The warm-up was extended to study the effect, and then the multi-speed, long slide-path,

and second multi-speed tests were performed in the order that they would be in all

subsequent tests. The results of the investigation is discussed further in the Appendix A.

Two tests, B2.15.999.100 and B2.15.999.005, were conducted on the 2B material to study

the effects of much higher (10 in/min) sliding speeds. After B2.15.999.100 had been

subjected to the complete regime of standard tests, an additional 48,000 inches of slide-

path were applied. This was done to obtain a better measurement of the rate of wear for

the interface.

21

4 Results

4.1 Observed Results

4.1.1 Temperature The temperature of the carrier plates at the sliding interface was measured with a laser

thermometer pointed at the PTFE carriers from the side. The measurements were made

while the test was running, so making contact measurements of the temperature of the

PTFE sliding surface itself was not possible.

The temperature at the carrier plates never varied more than 6º F from ambient during

any test in which temperature was measured. If any significant heat was generated by the

sliding, it was absorbed by the adjoining steel plates.

It was later discovered that changes in the ambient temperature affected the hydraulic

pressure in the normal force rams. This is discussed in Section 4.3.

4.1.2 Wear of PTFE The samples were photographed before and after the test. After the test, the #8 mirror

polish plates showed a coating of PTFE, and there were flakes of worn off PTFE around

the edges of the PTFE disks. An example is shown in Figure 4-1. The 2B rolled finish

plates (Figure 4-2) were visibly polished where they contacted PTFE. Flakes of PTFE

were evident on all used 2B specimens. Rough stainless steel plates (Figure 4-3) were

found to have been somewhat polished by the PTFE, and the PTFE appeared clean and

smooth. Flakes were not seen in the rough specimens.

22

Figure 4-1. mirror polished specimen after test M1.45.250.100

Figure 4-2. 2B rolled specimen after B2.45.250.100 test

23

Figure 4-3. Rough rolled specimen after R1.45.250.100 test

Observable wear seemed to correspond to the change in CoF experienced. #8 mirror

polished specimens transferred significant amounts of PTFE to the SS surface, and CoF

increased during the test. 2B rolled specimens lost some PTFE, and the SS surface

appeared polished, but was coated with less PTFE. The CoF rose less than it did for #8

specimens. Lost PTFE was not evident for rough rolled specimens, the SS surface became

polished, and no coating of PTFE was visible. CoF decreased for these specimens.

After the tests, the surfaces roughness of rough and 2B rolled specimens was measured, to

quantify the degree of polish, if any. The results of this measurement are shown in Table

4-1. At the time these measurements were taken, all the mirror finish plates had been re-

used as 2B backing plates, so no data is available for a polish effect on #8 mirror

specimens.

Table 4-1. Polishing effect of PTFE Stainless Steel interfaces.

Before (!in Ra)

After (_in Ra)

Polish Effect

Rough Rolled: 186.35 160.30833 13.97%2B Rolled: 5.1125 3.99375 21.88%

24

4.1.3 PTFE Loss After the first set of tests (R1.45.250.100 and M1.45.250.100) revealed some unexpected

behavior of rough stainless steel interfaces, the thickness of each specimen was measured

with a micrometer before and after each test, with the goal of obtaining data on loss of

PTFE by wear. Duplicate measurements on any one specimen were found to be

repeatable within 2 to 5 thousandths of an inch. Measurements of the loss of PTFE

thickness during the testing are given Table 4-2. In all cases the measurements imply some

loss of PTFE, presumably through transfer to the stainless steel and flaking off. However,

the magnitude of the loss is about the same as the precision of the measurement. Thus,

while they indicate that PTFE is being worn away, their usefulness for accurate prediction

of long term loss is questionable. Overall, the #8 mirror polish plates suffered PTFE

losses similar to those of the rough rolled specimens, and 2B rolled plates suffered very

little loss.

The PTFE loss data were collected mostly for tests at 1500 psi contact pressure, so

correlation of PTFE loss with contact pressure was not possible.

After the test of B2.15.999.100, the specimens were removed, measured, and re-installed

to run for 48000 inches more at 10 in/min and ±1.0” cyclic displacement amplitude.

When they were removed, as Table 4-2 shows, they had lost an additional 0.001 inches.

25

Table 4-2. Measurement of PTFE loss during the tests.

Specimen PTFE Loss (inches)

M1.15.384.005 0.002M1.15.250.100 0.006M1.30.250.100 0.005B2.15.250.100 0.002B2.30.250.100 0.000B2.45.250.100 -0.002B2.15.250.025 0.001B2.15.250.005 0.001B2.15.384.100 0.001B2.15.384.025 0.000B2.15.384.005 0.001B2.15.999.100 0.001B2.15.999.005 0.005R1.15.384.005 0.002R1.15.250.100 0.007R1.30.250.100 0.009

Mirror Plate Average: 0.0042B Plate Average: 0.001

Rough Plate Average: 0.006

Extreme Long Path Test:B2.15.999.100 0.001

4.2 Recorded Results

The data recorded during the test included:

• The horizontal actuator force and displacement

• The movements of the two static plates in the vertical and longitudinal directions,

• The hydraulic pressure in the rams providing contact pressure (except in tests

R1.45 and M1.45, where the force in the rams was measured via load cells on the

rods)

Data was collected automatically every 0.5 seconds in most tests. For the slowest tests, this

was reduced to once per second. For most of the long slide-path tests, two cycles out of

ten were recorded. This section shows examples of the raw data. The processed data are

shown and discussed in Chapter 5.

26

4.2.1 Load-displacement Curves Figure 4-4 shows some idealized test data. It has the following characteristics: a peak at the

“breakaway” point as the bearing begins to slide, and a plateau where friction is constant

(and a function of contact pressure and sliding speed). None of the curves generated in

this research show such behavior exactly.

Figure 4-4. Idealized force displacement curve.

Figure 4-5, Figure 4-6, and Figure 4-7 show load-displacement curves that exemplify most

of the behaviors observed in the testing.

27

Figure 4-5. Typical load-displacement curve for #8 mirror polish test

Figure 4-5 shows a load-displacement curve recorded during the warm-up to the multi-

speed test for Specimen M1.15.250. The sharp peaks at the upper left and lower right

show the static friction, or 'stiction', as the actuator reverses direction. The plot shows

fairly ideal behavior: small 'stiction' peaks at the beginning of each movement and a fairly

constant force thereafter. Note that the figure represents raw data and therefore shows

friction force, not friction coefficient. The vertical force varied slightly during each cycle.

Thus, when the coefficient of friction was obtained by dividing horizontal force by vertical

force, the curves generally became slightly smoother. The variations in friction force are

discussed in detail in Appendix A. The load-displacement curve is not exactly symmetric

about zero (The positive force is larger in magnitude than the negative force). This occurs

because of small offsets in the actuator. The average coefficient of friction during the loop

was determined from the loop area, as described in Section 4.3, so it was not affected by

this offset.

28

Figure 4-6. Typical load-displacement curve for 2B rolled plate test

Figure 4-6 depicts a load-displacement curve from the warm-up segment of a test on 2B

rolled stainless at 1500 PSI. It shows behavior typical of 2B rolled plates: very constant

friction during the stroke with a fairly small breakaway peak.

29

Figure 4-7. Typical load-displacement curve for rough rolled plate test

The load-displacement curve shown in Figure 4-7 was generated by a rough rolled

specimen, with 1500 psi contact pressure, and exemplifies the behavior typical of rough

rolled interfaces. Instead of a flat plateau, the friction gradually approaches a minimum

near the middle of the stroke, then climbs to a small peak just before the reversal point.

Some of this effect may also be due to variations in normal force with displacement,

which will be corrected in converting the actuator load to coefficient of friction.

A plausible explanation for this behavior is that the CoF of rough stainless steel is very

sensitive to slide-path. As shown in Section 5.3.4, the friction coefficient for this steel

decreases with slide-path. Every point on the PTFE sample goes through the same slide-

path. However, the middle of the plate, which is always in contact with moving PTFE,

experiences a longer slide-path than does the edge, and it therefore has the lowest friction

coefficient. A short test was done to confirm this hypothesis, in which the specimen

R1.45.250.100 was run out to a larger displacement than usual, after the standard series of

tests were finished. The trend continued for the larger displacements as predicted, see

Section 5.5. Appendix B shows a plot of CoF vs. slide-path for each specimen

30

4.3 Data Processing

The force-displacement curve for each test was first converted to a curve of coefficient of

friction vs. displacement. At each point, the horizontal force was divided by the

instantaneous vertical load, which was obtained from a digital pressure gage connected to

the four vertical load rams. The result was a value of the instantaneous coefficient of

friction at each recording time. Those curves appeared similar, but not identical, to the

force-displacement curves because the vertical load varied little during each cycle.

To compare results among specimens, the data were further processed and reduced to a

single value of friction for each cycle. This was done by computing the area inside the

coefficient of friction vs. displacement loop and dividing by the peak-to-peak

displacement. The area was found by numerical integration, using the trapezoidal rule.

This procedure results in an average friction coefficient for the conditions of that

particular cycle, and allows comparisons with other conditions to be made. Those

comparisons are made in Chapter 5.

Figure 4-8. Load displacement curve for B2.15.999.100 test showing individual datapoints.

31

Figure 4-8 shows an example of the individual data points used in analysis. Each of these

points represents one reading of actuator force and displacement, as well as hydraulic

pressure, pot displacements, etc. The actuator force is first converted to coefficient of

friction by dividing by the contact pressure, which is a function of the bearing

compression ram pressure. The result is a CoF-displacement curve. To compute the

average CoF over the cycle, the area contained by the CoF-displacement curve was

divided by the total displacement traveled during one cycle, as measured by the LVDT

attached to the MTS actuator (four times the cyclic displacement amplitude). This test was

performed at high speed (10 in/min), and so the data-points are somewhat sparse. This

resolution is sufficient to calculate the average friction, but the exact breakaway peak is

likely not captured. Most tests had much lower speeds, and better resolution of the data.

Some tests could not be analyzed in the standard manner because the sampling rate was

insufficient to define the loop well. (This occurred primarily during tests with small

displacement amplitudes and high frequency). Figure 4-9 shows an example. The

displacement should reach 0.05” on either side, but of the six data points recorded per

cycle, the largest had an absolute amplitude of only 0.036”.

Figure 4-9. Sparse data during the B2.15.384.005 does not adequately represent the cycle.

32

This data sparsity occurred in tests where the ratio of sliding speed to cyclic displacement

amplitude exceeded 10, which included M1.15.384.005, B2.15.384.005, R1.15.384.005,

B2.15.250.005, B2.15.384.025, B2.15.999.025 and B2.15.999.005. For these tests, data

points from 100 consecutive cycles were sorted by their location on the cyclic path, and

treated as one cycle. Figure 4-10 shows an example of this procedure, in which the data

from 100 cycles, all distributed randomly but superimposed on the same plot, provide

better definition of the loop. The analysis was then completed as usual. The method was

verified using borderline cases (where the speed/displacement ratio was between 5 and

10). Both methods produced very similar results. The method is viable if the CoF does not

change much over the 100 cycles used to create one composite cycle. Fortunately, the

problem occurred only in tests in which the cyclic amplitude was small, which in turn led

to very small changes in CoF per cycle.

Figure 4-10. Several deficient cycles superimposed reveal a better picture.

The loading and unloading lines on either side of the composite cycle are slightly sloped,

which reflects a small amount of elastic deformation of the PTFE. If this deformation is

excessive in relation to the displacement of the test, friction would be under-calculated for

33

small displacement tests. Section 5.3.5 explores this issue in detail. It was found to have a

negligible effect on the results.

34

35

5 Analysis & Discussion of Results

5.1 Introduction

Campbell and Kong (1989) identified the following variables as having an effect on

friction coefficient:

• Contact pressure • Speed of travel • Roughness of the mating surface • Length of the travel path • Dimpling and lubrication • Eccentric loading • Temperature • Creep • Type of PTFE • Attachment of the PTFE to the backing plate • Surface contamination • Load and travel history • Specimen size • Wear

The tests described here were designed to address the first four of these effects (contact

pressure, sliding speed, surface finish and slide-path). Because each test conducted on a

specimen adds to its cumulative slide-path, there is an inherent difficulty in separating

slide-path effects from others. However, this chapter describes the analysis conducted on

the data to distinguish the effects of each parameter on the coefficient of friction (CoF).

The data were analyzed with the goal of creating an empirical model for CoF based on the

four effects tested:

(1)

Where:

!0 = the basic CoF for one of the three surface finish under reference conditions

f(v) = a function of sliding speed, equal to 1.0 under reference conditions

g(p) = a function of contact pressure, equal to 1.0 under reference conditions

36

h(s) = a function of slide-path length, equal to 1.0 under reference conditions

Reference conditions are taken as 3000 psi contact pressure, 2.5 in/min sliding speed and

zero slide-path. The terms of this equation will be defined as this chapter continues.

Multi-speed test results are described in Section 5.2:

• Section 5.2.1 investigates the effect of speed on the friction coefficient. In all tests,

CoF increased with sliding speed.

• Section 5.2.2 investigates the effect of contact pressure on the friction coefficient.

In all tests, CoF decreased with an increase in pressure until about 3000 psi. At

pressures higher than that, it remained essentially constant.

Long slide-path test results are described in Section 5.3

• Section 5.3.1 explains issues in processing data, and the methods employed in

correcting the data.

• Section 5.3.2 investigates the effect of speed on the friction coefficient. As in the

multi-speed tests, CoF increased with sliding speed.

• Section 5.3.3 investigates the effect of contact pressure on the friction coefficient.

Again, CoF decreased with an increase in pressure until 3000 psi, and was

approximately constant at higher pressures.

• Section 5.3.4 investigates the effect of slide-path on friction. In #8 mirror polish

specimens, friction increased with slide-path, in rough finish specimens, friction

decreased, and in 2B rolled specimens friction increased, but at a rate much lower

than that of #8 mirror polish.

• Section 5.3.5 investigates the effect of cyclic displacement amplitude (stroke) on

the rate of wear. The rate at which friction increased appeared not to change with

different stroke amplitudes.

Warm-up movement results are described in 5.4:

• Effect of slide-path on friction for very small slide-paths.

37

• Maximum expected breakaway friction.

Results of the extreme cyclic displacement amplitude test are described in 5.5.

Discussion of these results is in section 5.6.

5.2 Short Slide-Path, Multi-Speed, Tests

5.2.1 Effect of Sliding Speed Previous research shows that higher sliding speeds produce higher friction coefficients

(Constantinou et al., 1999). The trend is particularly important for sliding devices used for

seismic isolation, although those lie outside the scope of the work reported here. More

relevant to this investigation are the different sliding speeds to be expected under thermal

and traffic loading. In this section, the results of the multi-speed tests are used to study the

effect of sliding speed.

The multi-speed tests were performed in alternating order, with the fastest and slowest

tests done first, then the second-fastest and second-slowest, and so on. This order was

intended to separate the effects of slide-path from the effects of friction. Figure 5-1 shows

artificial CoF vs. sliding speed data. The plot on the left shows how data would look for

an interface for which CoF is increasing with slide-path. The plot on the right shows how

data would look for an interface for which CoF is decreasing with slide-path.

Figure 5-1. Example data showing data increasing or decreasing with slidepath.

38

In the plots of actual data, there is some evidence of both behaviors, but generally, the

CoF remains relatively close to the trendline. The implication is that the sequence of

testing in the multi-speed tests was not important after all. It was, however, a predunt

precaution.

The coefficients of friction from the multi-speed tests were plotted against sliding speed

for each finish of stainless steel in Figure 5-2, Figure 5-3 and Figure 5-4. The plots for

before and after the long slide-path test are shown side by side for ease of comparison. In

all cases, the friction coefficient used is the average over three cycles.

Figure 5-2. Multi-speed test, 1500 psi contact pressure, Long slide-path test at 2.5 in/min, ±1.0” stroke

39

Figure 5-3. Multi-speed test, 3000 psi contact pressure, Long slide-path test at 2.5 in/min, ±1.0” stroke

Figure 5-4. Multi-speed test, 4500 psi contact pressure, Long slide-path test at 2.5 in/min, ±1.0” stroke

Several trends are visible from the plots:

• In almost all cases, the CoF increases with sliding speed. The only exception was

with the 2B finish at 4500 psi before the long slide-path, when the friction

remained constant. However, only a few data points are available from those tests.

Contact pressure was found to be unacceptably low for the last several sliding

speeds, and those data were thrown out.

40

• The rough material displayed the highest friction in the tests run before the long

slide-path, and among the lowest after the long side-path. This result was

unexpected, but it occurred consistently for all rough specimens. Because the tests

on different materials were conducted in a somewhat random sequence, the result

cannot be a consequence of some unintended change in the test conditions.

• The imposition of a long slide-path loading caused the CoF for #8 mirror polished

specimens to rise, while the CoF for rough rolled specimens dropped. The CoF

for 2B rolled specimens rose, but less than that of the #8 mirror specimens.

• The 2B material showed a CoF that was in many cases less than that of the #8

mirror polish material.

Figure 5-5, Figure 5-6, and Figure 5-7 show the ratio of 2B CoF to #8 mirror polish CoF.

These are useful in determining the suitability of stainless steel with a 2B finish as a

substitute for a #8 mirror polished finish at different speeds. They were produced by

taking the ratio of the trend-lines, not the data points, because using the individual data

points would lead to large scatter in the data. The curves can be thought of as representing

the suitability of 2B finish stainless steel as a substitute for #8 mirror polished stainless

steel with respect to speed and slidepath. Where the ratio is less than 1.0, 2B steel has

superior performance. Where the ratio is larger than 1.0, 2B is inferior.

41

Figure 5-5. Multi-speed test at 1500 psi. Ratio of B2 friction to M1 friction vs. speed before and after the long slide-path test at 2.5in/min and ±1.0” stroke

Figure 5-6. Multi-speed test at 3000 psi. Ratio of B2 friction to M1 friction vs. speed before and after the long slide-path test at 2.5in/min and ±1.0” stroke

42

Figure 5-7. Multi-speed test at 4500 psi. Ratio of B2 friction to M1 friction vs. speed before and after the long slide-path test at 2.5in/min and ±1.0” stroke

As can be seen from Figures 5-5 thru 5-7, the CoF for 2B material is lower than that of #8

mirror polished material under all circumstances except a contact pressure of 4500 psi,

where 2B rises to about 1.6 times the friction of #8 mirror polished stainless steel.

In order to find an approximate linear function relating CoF to sliding speed for the

model in Equation 1, the results of the sliding speed tests from all tests were normalized

to the mean of the CoFs from the 2.5 in/min test, so that the factor at the reference

condition sliding speed is 1.0. The normalized data are presented in Figure 5-8, along with

trend-lines computed for each finish using a linear least squares method.

43

Figure 5-8. All multi-speed tests, before and after a long slide-path test, friction amplification factor normalized to 1 at 2.5in/min for each test.

Figure 5-8 shows that friction amplification factors from all three finishes of stainless steel

are about equally sensitive to sliding speed. #8 mirror polished and 2B rolled stainless steel

are particularly similar. The function is given by:

(2)

Where v is the sliding speed in in/min and the coefficients av and bv are given in Table 5-1.

Table 5-1. Coefficients for the amplification function of sliding speed

av bvmin/in

Mirror Polish 0.13 0.672B Rolled 0.13 0.68

Rough Rolled 0.12 0.71

Speed

As can be seen the constants av and bv are nearly the same for all finishes, suggesting that a

single pair of constants may provide an adequate representation of the sliding speed

sensitivity of all three material pairs.

44

5.2.2 Effect of Contact Pressure For each finish and each contact pressure, three points were taken from the multi-speed

test at the lowest speed (0.08 in/min) an intermediate speed (0.064 in/min) and the

highest speed, (2.5 in/min). Those points were then plotted against the contact pressure.

The plots are shown in Figure 5-9, Figure 5-10 and Figure 5-11

Figure 5-9. CoF vs. contact pressure for 0.08 in/min, before and after long slide-path test at 2.5 in/min and ±1.0” stroke

Figure 5-10. CoF vs. contact pressure for 0.64 in/min, before and after long slide-path test at 2.5 in/min and ±1.0” stroke

45

Figure 5-11. CoF vs. contact pressure for 2.5 in/min, before and after long slide-path test at 2.5 in/min and ±1.0” stroke

Figure 5-9, Figure 5-10 and Figure 5-11 show CoF vs. contact pressure at three speeds

from the multi-speed tests. The expected behavior is that CoF would be constant between

3000 and 4500 psi, and would increase at lower pressures, and this general trend can be

seen in the plots. However, there is too much scatter to make any clear conclusion about

the effects of contact pressure from these plots.

46

Figure 5-12. Ratio of B2 friction to M1 friction vs. contact pressure before and after the long slide-path test for 0.08 in/min.

Figure 5-13. Ratio of B2 friction to M1 friction vs. contact pressure before and after the long slide-path test for 0.64 in/min.

47

Figure 5-14. Ratio of B2 friction to M1 friction vs. contact pressure before and after the long slide-path test for 2.5 in/min.

Figure 5-12 and Figure 5-14 show the ratio of 2B CoF to Mirror CoF plotted against

contact pressure for different speeds. At low pressures, the ratio is less than one, meaning

that 2B finish specimens had lower CoFs. At 4500 psi, the CoF for 2B finish was always

equal to or higher than that for the mirror polish material.

In Figure 5-12, Figure 5-13, and Figure 5-14, some data points are missing. This is because

those data points were found to be spurious due to an accidental pressure drop, and so

were omitted (B2.45.250.100, before long slide-path), or escaped recording (due to

operator error in B2.45.250.100, after long slide-path). Fortunately, due to the order of the

tests, whereby the fastest and slowest tests are completed first, the unaffected data (before

the drop in pressure) are usually sufficient to describe the trend. In cases where the first

(fastest) cycles were not recorded, there is less certainty, as the next fastest speed was 1.25

in/min, half the speed of the first (2.5 in/min).

48

5.3 Long Slide-path Tests

5.3.1 Problems with Data and Correction Methods The resources available for the project did not extend to a servo-controlled hydraulic

system for maintaining constant pressure on the normal force rams. Thus a simpler system

was used, which resulted in some variation of the normal force during the long slide path

tests. It consisted of pumping up the system to the desired pressure, or slightly higher,

closing the hydraulic shutoff valves, and re-pumping the system back up to the desired

pressure if and when the pressure dropped unacceptably.

The variations in pressure were only detected from the very long slide-path test

(2B.15.99